Ore Geology Reviews 107 (2019) 30–40

Contents lists available at ScienceDirect

Ore Geology Reviews

journal homepage: www.elsevier.com/locate/oregeorev

In situ major and trace element analysis of magnetite from carbonatite- T

related complexes: Implications for petrogenesis and ore genesis
⁎
Wei Chena, , Yuan-Can Yinga, Tian Baia, Jin-Jun Zhanga, Shao-Yong Jianga, Kui-Dong Zhaoa,
Dongbok Shinb, Jindrich Kynickyc
a
State Key Laboratory of Geological Processes and Mineral Resources, Collaborative Innovation Center for Exploration of Strategic Mineral Resources, Faculty of Earth
Resources, China University of Geosciences, Wuhan 430074, China
b
Department of Geoenvironmental Sciences, Kongju National University, Gongju 314-701, South Korea
c
Department of Geology and Pedology, Mendel University in Brno, Zemedelska 3 61300 Brno, Czech Republic

A R T I C LE I N FO A B S T R A C T

Keywords: Magnetite (Fe3O4) is one of the most common accessory minerals in magmatic rocks, and it can accommodate a
Magnetite wide variety of major, minor and trace elements that can be measured by laser ablation ICP-MS. In this study, we
Magmatic investigate the chemical compositions of magnetite from four carbonatite complexes (Oka, Mushgai Khudag,
Hydrothermal Hongcheon and Bayan Obo). The minor elements (Mg, Ti, Al, Mn) in magnetite vary significantly both within
Carbonatite
and between different complexes. High field strength elements (Zr, Hf, Nb, Ta, U, Th) are generally depleted in
LA-ICP-MS
magnetite from carbonatite complexes, whereas K, Rb, Cs, Ca and P are commonly below detection limits. V and
Zn display significant variations from tens to thousands of ppm. Co, Ni and Ga are present in ppm or tens of ppm,
whereas Cu, Sr, Y, Ba and Pb are characterized by sub-ppm levels. Mo and Ge are identified at the ppm level,
whereas a consistent concentration of 2–5 ppm is observed for Ge. The determined chemical compositions of
magnetite from carbonatite complexes are quite distinguishable compared to those formed in silicate and sulfide
melts. This is clearly shown using multielement variation diagrams, and the distinct signatures of carbonatite-
related magnetite include strong positive anomalies of Mn and Zn and negative anomalies of Cu, Co and Ga. The
discriminant diagrams of Ti vs. Zr + Hf, Ti vs. Nb + Ta and Ni/Cr vs. Ti are applicable for distinguishing
magmatic and hydrothermal magnetite in carbonatite-related environments. In addition, the discriminant dia-
gram of Zn/Co vs. Cu/Mo and Cu vs. Zr + Hf can be used to distinguish carbonatite-related magnetite from
magnetite that formed in other environments.

1. Introduction ICP-MS have made it feasible to obtain a more comprehensive geo-

Magnetite (Fe3O4) is one of the most common accessory minerals in
magmatic rocks. It is a ubiquitous constituent of carbonatites and as-
sociated silicate and phosphate rocks (e.g., Reguir et al., 2008; Huang
et al., 2015; Ivanyuk et al., 2016; Milani et al., 2017). The structural
tolerance of magnetite to various substitutions means that it is able to
incorporate abundant major, minor and trace elements, such as Al, Ti,
Mg, Mn, Zn, Cr, V, Ni, Co and Ga. Importantly, magnetite has been used
as a petrogenetic and provenance indicator, as well as in the explora-
tion of ore deposits (Bailey and Kearns, 2002; Lee et al., 2005; Reguir
et al., 2008; Dare et al., 2014; Nadoll et al., 2014). Previous studies
have provided only limited trace element data due to the poor detection
limits of laser spectral analyses. Recent developments in laser ablation
chemical understanding of magnetite (e.g., Nadoll and Koenig, 2011;
Dare et al., 2012; Nadoll et al., 2014; Liu et al., 2015; Milani et al.,
2017).
The most recent studies have used trace elements in magnetite to
trace the conditions of ore formation, such as temperature, pressure,
oxygen fugacity, and cooling rate (e.g., Mollo et al., 2013; Nadoll et al.,
2014; Canil et al., 2016). A variety of elemental diagrams, or so-called
discriminant diagrams, have been developed to allow the use of mag-
netite as a petrogenetic indicator. Kamvong et al. (2007) suggested that
a Sn/Ga vs. Al/Co diagram can be used to discriminate different mag-
matic-hydrothermal ore deposits based on preliminary data. Dupuis and
Beaudoin (2011) proposed the use of a Si + Mg vs. Ni + Cr diagram for
the discrimination of magnetite in Ni-Cu-PGE sulfide deposits from

⁎
Corresponding author.
E-mail addresses: wchen@cug.edu.cn (W. Chen), shyjiang@cug.edu.cn (Y.-C. Ying), shyjiang@cug.edu.cn (S.-Y. Jiang), zhaokd@cug.edu.cn (K.-D. Zhao),
shin@kongju.ac.kr (D. Shin), jindrak@email.cz (J. Kynicky).

https://doi.org/10.1016/j.oregeorev.2019.01.029
Received 24 February 2018; Received in revised form 24 January 2019; Accepted 31 January 2019
Available online 10 February 2019
0169-1368/ © 2019 Elsevier B.V. All rights reserved.
W. Chen, et al.
hydrothermal and sedimentary environments and the use of an Al/
(Zn + Ca) vs. Cu/(Si + Ca) diagram to separate Cu-Zn-Pb volcanogenic
massive sulfide (VMS) deposits from other deposit types. Dare et al.
(2014) built a multielemental diagram that is normalized to the bulk
continental crust to discriminate magnetite crystallized in different
environments. Nadoll et al. (2015) presented a covariation diagram of
Al, Mn, Ti and V to discriminate porphyry from skarn magnetite.
Previous research of magnetite from carbonatites has mainly fo-
cused on the variation characteristics of major elements. There is a lack
of the systematic and complete analyses of the trace element char-
acteristics of magnetite from carbonatite (e.g., Prins, 1972; Mcmahon
and Haggertry, 1982; Gaspar and Wyllie, 1983; Ramasamy et al., 2001;
Bailey and Kearns, 2002; Reguir et al., 2008; Shin et al., 2013; Ivanyuk
et al., 2016). Mg, Al, Mn and Ti are the most abundant and variable
elements in carbonatite magnetite, and the depletion of Ta, Zr and Hf is
typical for the trace element geochemistry of magnetite from carbo-
natite (Ramasamy et al., 2001; Bailey and Kearns, 2002; Reguir et al.,
2008). Reguir et al. (2008) suggested that the presence of two or more
morphologically and chemically distinct types of magnetite within
carbonatite can provide an informative record of their evolution. For
instance, magnetite crystallizing from carbonatite magma evolves by
becoming depleted in Mg and Ti (Reguir et al., 2008). More recently,
the comprehensive chemical compositions of magnetite from carbona-
tite-related deposits have also been used to trace the origin of mineral
deposits and to distinguish different ore-forming processes (Huang
et al., 2015; Milani et al., 2017). Magnetites from the Bayan Obo
banded iron ores are characterized by Al, Mn, Ti, and V contents similar
to those of banded iron formations (BIF), whereas the Fe-oxides from
disseminated and massive ores show similar chemistry as those from
skarn and magmatic-hydrothermal deposits (Huang et al., 2015). The
most recent investigation of magnetite from the Phalaborwa carbona-
tite and phoscorite provided its full elemental composition and con-
firmed that magnetite is a good petrogenetic indicator (Milani et al.,
2017).
In this study, we used LA-ICP-MS to determine the major, minor,
and trace element compositions of magnetite from several typical car-
bonatite complexes, which will contribute to the characterization of
magnetite chemistry from different settings. The aim of this study is to
identify elements in magnetite indicating carbonatite host rocks, and
construct discriminant diagrams that separate carbonatite magnetite
from those from other origins.
2. Geological background and sample description
Magnetite-bearing samples from several carbonatite complexes
(Fig. 1) were analyzed, including the Oka carbonatite complex (Ca-
nada), the Mushgai Khudag carbonatite complex (Mongolia), the
Hongcheon carbonatite-phoscorite complex (South Korea), and the
Bayan Obo rare earth element (REE)-Fe-Nb deposit (China).
The Oka carbonatite complex is one of the westernmost intrusions of
the Monteregian Igneous Province (MIP), Canada. This complex con-
sists of both carbonatite and silicate rocks, including okaite, ijolite,
alnöite and jacupirangite (Fig. 1b; Chen and Simonetti, 2013, 2014;
Chen et al., 2013a). The carbonatite is calciocarbonatite in composi-
tion, and its accessory minerals include biotite, nepheline, melilite,
pyrochlore, perovskite, magnetite, sulfide and REE minerals (Chen and
Simonetti, 2013, 2014). The okaite consists of dominant melilite with
minor biotite, calcite, apatite, magnetite and perovskite (Chen and
Simonetti, 2013, 2014). Magnetite is a common magmatic accessory
mineral within the carbonatite and okaite from Oka, and it usually
contains abundant melt inclusions (its solid phases include carbonates,
halides, sulphates, sulfides and an admixture of phosphates, oxides and
silicates), which have been documented as being similar to the mineral
assemblages from the Oldoinyo Lengai natrocarbonatite lavas (Chen
et al., 2013a). Magnetite within the carbonatite usually occurs along
calcite boundaries and varies in grain size, whereas those within the
31
Ore Geology Reviews 107 (2019) 30–40
okaite are usually associated with melilite and show spinel exsolution
(Fig. 2a–c). Minor fine-grained pyrrhotite is observed disseminated or
on the edge of the magnetite within the okaite (Fig. 2b). Both the
carbonatite and okaite outcrop samples were collected during the 1986
GAC-MAC field trip (Gold et al., 1986). The magnetite from Oka has
been investigated as an example of typical magmatic magnetite from a
carbonatite-related environment.
The Mushgai Khudag (MK) subvolcanic-plutonic complex in the
Southern Gobi of Mongolia is composed of nelsonite, carbonatite and
syenitic varieties (Fig. 1c). It is the second largest REE-Fe-P-F-Sr-Ba
deposit in Mongolia (Baatar et al., 2013; Enkhbayar et al., 2016). The
Mushgai Khudag complex is controlled by E-W-stretching faults, and it
formed in a continental rift environment. Carbonatite usually occurs as
veins within the nelsonite and syenite, or it intrudes into brecciated
rocks (Baatar et al., 2013; Enkhbayar et al., 2016). The carbonatite
mainly consists of calcite, minor apatite and fluorite, and its accessory
minerals include magnetite, celestite (SrSO4), barite, quartz and REE
minerals (e.g., bastnaesite-(Ce), synchysite-(Ce), Ce- and Nd-parisite).
The nelsonite is dominantly composed of apatite and magnetite, with
accessory minerals including fluorite, barite, celestine, fluorcarbonates
and quartz. Magnetite within Mushgai Khudag occurs as large crystals,
with grain sizes of up to tens of centimeters, and it usually displays
ilmenite lamellae (Fig. 2d–f). Ilmenite exsolution in Ti-rich magnetite is
common in magmatic magnetite (e.g., Fe-Ti oxide-bearing mafic–ul-
tramafic intrusions) due to its high Ti content and formation tempera-
tures (Ray and Webster, 2007; Tan et al., 2016). Ilmenite exsolution can
be the result of “oxy-exsolution” (at > 600 °C), the vacancy relaxation
of cation-deficient solid solution or the subsolidus re-equilibration of
coexisting Fe-Ti oxides (Buddington and Lindsley, 1964; Lattard, 1995;
Tan et al., 2016). Ilmenite and celestine inclusions can be identified in
the Mushgai Khudag magnetite (Fig. 2e,f). Outcrop samples were col-
lected, and the Mushgai Khudag magnetite is used as another example
of magmatic magnetite that formed in a carbonatite-related environ-
ment.
The Hongcheon carbonatite complex is the most important Fe-REE
deposit in South Korea. The Hongcheon area is located at the northern
part of the Gyeonggi massif in the Korean Peninsula, which is one of the
most important Precambrian Fe-mineralization zones in South Korea
(Kim et al., 2005, 2016). The Gyeonggi massif mainly consists of Pa-
leoproterozoic basement gneisses and overlying supracrustal sequences.
The orebodies are mainly hosted in carbonatite dikes, which have in-
truded Precambrian biotite and granitic gneisses (Fig. 1a; Kim et al.,
2005). The Hongcheon carbonatite is composed of dominant calcite,
dolomite and ankerite, and auxiliary minerals including magnetite,
monazite, strontianite (SrCO3), quartz, barite and sulfide. The presence
of abundant Fe-carbonate and quartz associated with magnetite in the
observed carbonatite indicates their formation during the late hydro-
thermal stage (Shin et al., 2013). The anhedral magnetite within the
Hongcheon carbonatite complex occurs along the calcite boundary, and
pyrite is a common component of this mineral association (Fig. 2g).
Hongcheon magnetite serves as an example of hydrothermal magnetite
in a carbonatite-related environment.
The Bayan Obo REE-Fe-Nb ore deposit in Inner Mongolia, North
China, is the largest light REE deposit in the world and a substantial
source of iron and niobium (Fig. 1d; Smith et al., 2015; Fan et al., 2016;
Yang et al., 2017). It is located in the northern margin of the North
China Craton (NCC), along the Central Asian Orogenic Belt (Fan et al.,
2016). The basement rocks are composed of Neoarchean mylonitic
granite-gneiss, Paleoproterozoic syenite and granodiorite, and biotite
granite-gneiss (Yang et al., 2017). The giant REE-Fe-Nb orebodies are
mainly hosted in dolomite marble (H8). The genesis of this giant de-
posit is still controversial, with a debate over whether it is sedimentary
or carbonatite in origin, although Devonian metasomatism is common
and has been identified in the H8 dolomite marble (Smith et al., 2015;
Fan et al., 2016; Yang et al., 2017). The dolomite marble sample in this
study was collected from the drill core in the East Orebody at a depth of
W. Chen, et al.
Fig. 1. Simplified geological maps of the investigated complexes, including Hongcheon (a), Oka (b), Mushgai Khudag (c) and Bayan Obo (d) adapted from Kim et al.
(2005), Chen and Simonetti (2013), Baatar et al. (2013), and Fan et al. (2016), respectively.
1540–1720 m in order to avoid surface alteration (Fig. 1d; Song et al.,
2018). Magnetite is a common accessory mineral in the dolomite
marble associated with REE minerals (e.g., bastnaesite, monazite), and
the latter was mostly characterized by a Devonian age of approximately
450 Ma (Smith et al., 2015; Song et al., 2018). The magnetite varies in
grain size and usually occurs as anhedral crystals (Fig. 2h). Based on the
previous investigations of Bayan Obo, which is believed to have formed
through extensive metasomatism (Smith et al., 2015; Fan et al., 2016;
Song et al., 2018; Yang et al., 2017), the magnetite from the dolomite
marble serves as an example of carbonatite-related hydrothermal
magnetite.
3. Analytical methods
Electron microprobe analysis (EMPA) is a conventional method used
to measure major and some trace elements in magnetite, but many trace
elements (e.g., Co, Cr) in magnetite are close to or below EMPA de-
tection limits; thus, elemental data obtained using EMPA are limited.
Recently, LA-ICP-MS has become the most popular method used for the
determination of the minor and trace elements in magnetite (e.g.,
Savard et al., 2010; Nadoll and Koenig, 2011; Dare et al., 2012, 2014;
Liu et al., 2015; Canil et al., 2016; Milani et al., 2017). In this study, the
major and trace element concentrations of magnetite were determined
by a Resonetics RESOlution S155-LR 193 nm laser ablation system
coupled to a Thermo iCAP-Q ICP-MS at the State Key Laboratory of
Geological Processes and Mineral Resources, China University of
Geosciences (Wuhan). The full list of the analyzed isotopes using LA-
ICP-MS includes 7Li, 23Na, 25Mg, 27Al, 29Si, 31P, 39K, 42Ca, 45Sc, 47Ti,
32
Ore Geology Reviews 107 (2019) 30–40
51
V, 53Cr, 55Mn, 57Fe, 59Co, 60Ni, 63Cu, 66Zn, 71Ga, 73Ge, 85Rb, 88Sr, 89Y,
90
Zr, 93Nb, 95Mo, 115In, 118Sn, 133Cs, 137Ba, 139La, 140Ce, 141Pr, 146Nd,
147
Sm, 153Eu, 157Gd, 159Tb, 163Dy, 165Ho, 166Er, 169Tm, 173Yb, 175Lu,
178
Hf, 181Ta, 182W, 208Pb, 232Th, and 238U. The ablation of the samples
was performed using helium as the carrier gas. This was mixed in a
conical ablation cup with the argon sample gas of the ICP-MS and ni-
trogen as the make-up gas. A small amount of nitrogen (i.e.,
4 ml min−1) was added to the central gas flow to increase the sensitivity
for most elements by a factor of 2–3 (Hu et al., 2008). Each spot ana-
lysis included approximately 30 s of background acquisition and 40 s of
sample data acquisition. A spot size of 33 μm was used together with a
repetition rate of 10 Hz and an energy density of 5–8 J cm−2. For the
Mushgai Khudag magnetite, raster mode with a spot size of 50 μm,
repetition rate of 6 Hz, energy density of 6 J cm−2, and scan speed of
5 μm s−1 was adopted to obtain the chemical composition of the mag-
netite-ilmenite solid solution. In general, 8–10 magnetite grains with
different petrographic textures were analyzed. If they displayed large
variations, an additional 8–10 magnetite grains were analyzed to obtain
representative results.
Previous analytical procedures usually require the use of the major
element Fe as the internal standard for the elemental concentration
characterization using LA-ICP-MS (Nadoll and Koenig, 2011; Dare
et al., 2012, 2014; Liu et al., 2015; Canil et al., 2016; Milani et al.,
2017). Commonly, this requires a previous determination of the Fe
content, which is usually determined by electron microprobe. Alter-
natively, Milani et al. (2017) used the 57Fe stoichiometric concentration
of 72.4 wt% in magnetite for the calibration. This assumption is only
valid when the total concentration of minor elements is < 3 wt%,
W. Chen, et al. Ore Geology Reviews 107 (2019) 30–40

Fig. 2. Petrographic images of investigated magne-

tite samples. (a) Magnetite within the Oka carbona-
tite, together with calcite (cal), melilite (mel) and
apatite (ap); (b) magnetite from the Oka okaite con-
tains abundant inclusions and is usually associated
with melilite and pyrrhotite (po); spinel exsolution is
commonly identified in the okaite, as shown in the
backscattered electron (BSE) image (c), and sec-
ondary mineral phases, including Na-enriched car-
bonate (carb) and perovskite (pv), are shown in the
inclusions; (d) coarse-grained magnetite, associated
with apatite and pyrrhotite, can be identified in the
Mushgai Khudag nelsonite, and ilmenite (ilm) ex-
solution lamellae are common in these magnetite, as
shown in the BSE images (e,f); Mushgai Khudag
magnetite usually contains ilmenite and celestine
(cls) inclusions (e,f); anhedral magnetite embedded
in calcite and dolomite (dol) are shown in samples
from Hongcheon (g) and Bayan Obo (h), respectively.

i i
which will result in a < 5 wt% difference in the Fe values measured Crm = concentration of element i in the reference material, cpssam = net
i
(Milani et al., 2017). In this study, we obtained the concentrations of 50 count rate of element i in the sample, cpsrm = net count rate of element i
is
elements, using multiple reference materials (NIST 610, BIR-1G, BCR- in the reference material, cpssam = net count rate of internal standard
is
2G and BHVO-2G) as external standards. The elemental concentration element in the sample, cpsrm = net count rate of internal standard ele-
is
calibration is similar to the internal standard-independent calibration ment in the reference material, Csam = concentration of internal stan-
is
strategy for the LA-ICP-MS analyses of anhydrous minerals proposed by dard element in the sample, and Crm = concentration of internal stan-
Liu et al. (2008). In short, the calibration involves both external cali- dard element in the reference material. For a given anhydrous mineral,
bration against the above reference glasses and internal standardiza- such as magnetite, the sum of all concentrations expressed as oxides
tion, as demonstrated by the following Eq. (1): should be 100 wt%. With this quantification, the concentration of the
element i can be determined with the following Eq. (2):
cps i ⎞ ⎛ cpsrmis is
⎞ ∗ ⎜⎛ Csam ⎞⎟
i
Csam i
= Crm ∗ ⎜⎛ sam i ⎟
∗ ⎜ is ⎟ is
⎝ cpsrm ⎠ ⎝ cpssam ⎠ ⎝ Crm ⎠ (1)
i
where Csam = concentration of element i in the sample,

33
W. Chen, et al.
Fig. 3. Cross-comparison of the EMPA and LA-ICP-MS analytical data for Mg, Al, Ti, Mn, Zn (a) and Fe (b) in magnetite. The EMPA data for magnetite can be found in
Chen et al. (2013a).
( )
i
i crm
100 ∗ cpssam ∗ i
i
Csam =
cpsrm
j Hongcheon (Supplementary Table 1). The Ti contents in magnetite
N c
∑ j = 1 ⎛cpssam
⎜
j
∗ ⎛ rmj ⎞ ⎞ ⎟
⎝ ⎝ cpsrm ⎠ ⎠
The off-line selection and integration of background and analyte
signals, time drift correction and quantitative calibration were per-
formed by ICPMSDataCal 10.2 (Liu et al., 2008). The time-resolved
analytical signals for each magnetite analysis were carefully checked.
Acquisitions with anomalous signal changes in the time-resolved signal
spectra, which indicate the ablation of a different phase, were discarded
(i.e., inclusions). Usually, anomalous increases in the intensities of Na,
Si, Ca, P, Mg, Nb, Zr, Sr, REE, Y, Th and U in the time-resolved signal
spectra are observed.
The calibrated major and minor element compositions (e.g., Mg, Al,
Ti, Mn, Zn and Fe) analyzed by LA-ICP-MS are consistent with the
EMPA data (Fig. 3; Chen et al., 2013a). For instance, all of the minor
elements in magnetite (i.e., Mg, Al, Ti, Mn and Zn) show very good
linear correlations between the two analytical techniques in the cross-
comparison diagram (Fig. 3a). The iron content in magnetite de-
termined by LA-ICP-MS is also consistent with the EMPA-obtained data,
as shown in the cross-comparison plot (Fig. 3b).
Both trace element contents characterized with an internal standard
(i.e., Fe) and characterized with multiple external references without an
internal standard were calculated. For instance, the Fe concentration of
the Oka magnetite was previously determined by EMPA (Chen et al.,
2013a); thus, its trace element compositions were calibrated both with
and without an internal standard. The relative deviations of the con-
centrations for all determined elements using both methods are < 10%
(Fig. 4). Thus, we suggest that the chemical (major, minor and trace)
compositions of magnetite can be obtained accurately using LA-ICP-MS
by adopting multiple reference materials for external calibration. The
major, minor and trace element data for magnetite from a selected
number of carbonatite complexes, determined using LA-ICP-MS, are
listed in Supplementary Tables 1 and 2.
4. Results
4.1. Major and minor element variations in magnetite
Magnetite from Bayan Obo and Hongcheon are characterized by
higher FeO contents (∼92.7 wt%) compared with those from Mushgai
Khudag and Oka, which vary from 70.6 to 90.9 wt%. By contrast, the
Ore Geology Reviews 107 (2019) 30–40
minor element concentrations (e.g., Ti, Al, Mg, Mn) in Mushgai Khudag
and Oka magnetite are much higher than those from Bayan Obo and
(2) from Mushgai Khudag and Oka generally exceed 10,000 ppm, but those
in the Bayan Obo and Hongcheon magnetite are lower (< 500 ppm).
The magnetite from the Oka okaite is characterized by the highest Al
contents (1.70–4.22 wt%) out of all groups studied and shows spinel
exsolution in the backscattered electron image (Fig. 2c). Lines/spots
were ablated across a magnetite grain characterized by spinel exsolu-
tion in order to determine the composition of the primary magnetite
prior to spinel exsolution. The Al abundance in the magnetite from
Bayan Obo, Hongcheon and Mushgai Khudag is relatively low
(< 5000 ppm). Extremely high Mn contents are identified in Oka
magnetite (up to 6.77 wt%), and the magnetite from Mushgai Khudag
also contains high Mn abundances (average ∼2000 ppm). The Mn
concentrations for the Bayan Obo and Hongcheon magnetite are gen-
erally below 1000 ppm. The highest Mg content (up to 2.84 wt%) is also
identified in magnetite from the Oka okaite. The Mg abundances for
magnetite from Bayan Obo, Mushgai Khudag and Hongcheon are re-
latively low (e.g., < 5000 ppm), whereas the magnetite from the Oka
carbonatite contains highly variable Mg contents (0.07–2.43 wt%). The
concentration of Mg displays a positive correlation with Al for all in-
vestigated magnetite (Fig. 5a), and this correlation has been observed in
magnetite from other carbonatites (e.g., Reguir et al., 2008). Spinel,
ulvospinel, jacobsite and magnesioferrite solid solutions are responsible
for the elevated concentrations of Mn, Mg and Al in Oka magnetite
(Reguir et al., 2008). Most of the SiO2 contents are between 0.2 and
0.4 wt%. It is known that the Si contents determined by LA-ICP-MS are
affected by high background intensity caused by the analysis of silicate
mineral phases, which could possibly cause the overestimation of the Si
composition (e.g., Dare et al., 2014; Milani et al., 2017). Na, K, Ca and P
are commonly close to or below detection limits in LA-ICP-MS, except
for the few hundreds of ppm of phosphorus that have been identified in
magnetite from the Mushgai Khudag nelsonite (Supplementary
Table 1).
4.2. High field strength element variations in magnetite
High field strength elements (HFSEs) are moderately compatible in
magnetite. However, these elements are measured at ppm or sub-ppm
levels in magnetite from carbonatite complexes (Supplementary
Table 1). Scandium is generally at the ppm level in magnetite from
carbonatite-related alkaline rocks (Supplementary Table 1). Zirconium
34
W. Chen, et al. Ore Geology Reviews 107 (2019) 30–40

Fig. 4. Relative deviations of element concentrations from multiple reference calibrations relative to internal standard calibration in magnetite. Relative deviation is
defined as 100*(measured values-preferred values)/preferred values of element concentrations in the analyzed magnetite.

is normally close to detection limits in Hongcheon and Bayan Obo
magnetite, whereas those from Oka and Mushgai Khudag contain much
higher contents of up to 98.7 ppm. Oka magnetite contains the highest
concentrations of Nb (∼37.6 ppm) and Ta (∼3.91 ppm), whereas these
metals within other magnetite groups are close to or below detection
limits, with the exception of the Mushgai Khudag magnetite, which
contains ∼ 0.7 ppm of Ta (Supplementary Table 1). The average Hf
concentration for magnetite from Mushgai Khudag and Oka is
0.69 ppm, whereas it is close to detection limits for magnetite from
Hongcheon and Bayan Obo. The thorium concentrations vary from
below detection limits up to 0.540 ppm in magnetite from carbonatite
related rocks (Supplementary Table 1).
literature since they are usually below detection limits (e.g., Nadoll
et al., 2015). Despite their high incompatibility in igneous Fe-oxide,
REEs are still detectable in some magnetite from carbonatite complexes
(e.g., Mushgai Khudag). This is because carbonatite melts and associate
alkaline silicate melts are extremely enriched in REEs (e.g., 3.1 wt%
REO for the Bear Lodge carbonatite), especially in light REEs (LREEs;
e.g., (La/Yb)CN = 485 for Bear Lodge carbonatite; Chakhmouradian
and Zaitsev, 2012). Only magnetite from Mushgai Khudag has a sig-
nificant REE content of 1.95–18.4 ppm, whereas the rest of the mag-
netite contains 0.006–1.21 ppm REE. The magnetite from the Mushgai
Khudag complex displays a weakly LREE-enriched pattern in the
chondrite-normalized REE diagram (Fig. 6).

4.3. Rare earth element variations in magnetite 4.4. Redox-sensitive element and other trace element variations in magnetite

Rare earth elements (REEs) are rarely reported for magnetite in the Uranium is at sub-ppm quantities (0.001–0.225 ppm) in most

Fig. 5. Covariation of elements in magnetite from the carbonatite complexes (a) Al vs. Mg; (b) Al vs. Ga; (c) Zn vs. Mn; (d) Cu vs. Zn.

35
W. Chen, et al.
Fig. 6. Chondrite-normalized REE patterns in magnetite from the carbonatite
complexes. Chondrite values are taken from McDonough and Sun (1995).
magnetite from the carbonatite rocks, whereas ppm-level concentra-
tions of U, with an average content of 1.43 ppm, have been identified
for several magnetite grains from the Mushgai Khudag nelsonite
(Supplementary Table 1). Vanadium displays a huge variation for dif-
ferent complexes, varying from tens (Hongcheon) to hundreds (Bayan
Obo) to thousands (Oka and Mushgai Khudag) of ppm. Chromium is
mostly at ppm quantities, with an average concentration of 3.59 ppm in
magnetite from Hongcheon and Oka, whereas the Cr contents in mag-
netite from Bayan Obo vary from 0.49 to 41.5 ppm, and the Cr content
in magnetite from Mushgai Khudag reaches an average concentration of
110 ppm. This, in general, agrees with the previous observation that the
Cr content is low in magnetite from carbonatite and related silicate
rocks (Bailey and Kearns, 2002; Reguir et al., 2008; Guzmics et al.,
2011).
Cobalt, nickel and gallium are present at the ppm or tens of ppm
level, with average concentrations of 42.6 ppm, 30.6 ppm, and
20.6 ppm, respectively, whereas Zn varies from tens to thousands of
ppm, with an average content of 1287 ppm. Gallium correlates well
with Al (Fig. 5b) for magnetite from all the investigated complexes.
Zinc displays a positive correlation with Mn in all studied magnetite
(Fig. 5c). The concentration of Ge is consistently ∼2–5 ppm. Magnetite
from the Mushgai Khudag nelsonite contains variable contents of Cu,
Sr, Y, Ba and Pb that vary from several ppm to hundreds of ppm,
whereas the magnetites from Bayan Obo, Hongcheon and Oka consist of
sub-ppm quantities of these given elements. Cu shows a correlation
with Zn in Fig. 5d. Except for the magnetite from Bayan Obo, which
contains a few ppm of Mo.
4.5. Trace elements in magnetite with exsolution textures
The magnetite from the Oka okaite shows spinel exsolution.
However, the exsolution is very small and relatively homogeneously
distributed (Fig. 2c). Thus, chemical zonation is not observed for the
magnetite from the okaite. The Mushgai Khudag magnetite displays
chemical zonation with respect to ilmenite exsolution textures. As
shown in Fig. 7 and Supplementary Table 1, the ilmenite lamellae are
characterized by increased concentrations of Al, Ti, Ge, Sn, Zr, Hf, Nb,
Ta and are depleted in V and Zn compared to the magnetite. The
composition of magnetite including the ilmenite lamellae represents the
primary magmatic titanomagnetite composition, whereas the spot
analysis represents the magnetite composition after ilmenite exsolution
during cooling. It is worth mentioning that the titanomagnetite displays
higher REE enrichments compared to the magnetite component (Fig. 6).
36
Ore Geology Reviews 107 (2019) 30–40
5. Discussion
5.1. Magnetite chemistry as a petrogenetic indicator
Dare et al. (2014) used a multielement variation diagram that
provides an overview of the concentration of a wide range of elements
within magnetite. The magnetite chemical data are normalized to the
bulk continental crust, which is closer in composition to an evolved
silicate melt. For the mantle-derived carbonatite rocks, the primitive
mantle might serve as a more reasonable source for the normalization.
However, to make comparisons to the much more widely distributed
silicate and sulfide melts in which magnetite forms, we use the bulk
continental crust composition (Rudnick and Gao, 2003) as the nor-
malization factor in Fig. 8. Magmatic magnetite from the Fe-Ti-V-P
deposits is used as a comparison to the magmatic magnetite from the
carbonatite complexes since both of them are associated with plume-
related magmas in a rift environment and commonly form from evolved
silicate melt that is either ferrodiorite or nephelinite in composition
(Barnes et al., 2004; Bell and Simonetti, 2010; Chen et al., 2013b).
Magmatic magnetites from carbonatite environments (i.e., Oka and
Mushgai Khudag) are generally enriched in Ge, Mo, Sn and depleted in
Cu, Ni and Cr compared to the primitive magnetite within Fe-Ti-V de-
posits (Fig. 8a). In addition, the carbonatite-related magmatic magne-
tites from Oka and Mushgai Khudag display positive anomalies of Ge,
Nb, Ta, Sn, Mn, Zn and V and negative anomalies of Cu, Ga, Mg and Co.
Positive Nb, Ta and Zn anomalies and negative Cu anomaly are char-
acteristic of carbonatite-related magmatic magnetite in the continental
crust-normalized diagram (Fig. 8a). The chemical signatures of the
hydrothermal magnetite from Bayan Obo and Hongcheon, in general,
resemble those of magnetites from silicate and sulfide systems (Fig. 8b;
Dare et al., 2012). The carbonatite-related hydrothermal magnetite is
depleted in Zr, Hf and Cu and enriched in P compared to other hy-
drothermal magnetite, whereas it displays positive anomalies of P, Ge,
Nb, Sn, Mn and Zn and negative anomalies of Zr, W, Cu, Ga, Mg and Co.
The negative W and Cu anomalies and positive Zn anomaly are char-
acteristic of the carbonatite-related hydrothermal magnetite (Fig. 8b).
Thus, these data support that the multielement diagram is useful for
discriminating magnetite from different geological settings, especially
for those from carbonatite-related complexes. As shown in Fig. 8, both
the magmatic and hydrothermal magnetite from the carbonatite com-
plexes generally show distinct anomalies in their normalized diagrams,
i.e., strong positive anomalies of Zn and Mn and negative anomalies of
Cu, Ga and Co.
These anomalies observed in the multielement variation diagrams
may record the chemical signature of the magma source, the partition
coefficients of the elements in magnetite, and/or the competition for
elements between phases cocrystallizing with magnetite (e.g., ilmenite
and sulfide). Therefore, the multielement characteristics of magnetite
can serve as a good petrogenetic indictor. For instance, the depletion of
Cu in magnetite relative to other incompatible elements, such as Nb and
Mo, recorded in Fe-Ti-P samples is caused by the cocrystallization of
magnetite and sulfide (Dare et al., 2014). The depletion of Cu is also
distinctively identified in magmatic magnetite formed in carbonatite-
related environments (Fig. 8), whereas the cocrystallization of sulfide
and magnetite is not commonly observed, or occurs in minor quantities,
within magmatic carbonatite and associated silicate rocks when melts
are depleted in sulfur (e.g., below detection limits for Oka samples;
Gold et al., 1986). The Cu contents in the Oka carbonatite vary from 5
to 30 ppm, with an average value of 14.4 ppm (Gold et al., 1986), which
is lower than that of the bulk continental crust (28 ppm; Rudnick and
Gao, 2003). Thus, the negative anomaly of Cu most likely records the
signature of the carbonatite melt and/or the partition coefficient of Cu
between magmatic magnetite and the melt. For the hydrothermal
magnetite from Hongcheon and Bayan Obo, the negative Cu anomaly
may record the copper depletion of both the carbonatite magmatic
source and the hydrothermal fluid and was possibly accompanied by
W. Chen, et al.
Fig. 7. Time-resolved analysis (TRA) spectra for LA-ICP-MS raster analysis of the Mushgai Khudag magnetite with ilmenite lamellae.
Fig. 8. Multielement variation diagrams of magnetite from different genetic
types normalized to the bulk continental crust (values are from Rudnick and
Gao, 2003). Average chemical compositions of magnetite from the carbonatite
complexes are plotted with magmatic magnetite from Fe-Ti-V deposit (a) and
hydrothermal magnetite from typically hydrothermal deposits (b). Data refer to
Dare et al. (2014).
the cocrystallization of sulfide in the case of the Hongcheon magnetite
(Fig. 2). Unfortunately, no Nb and Mo contents are available for the
bulk carbonatite of the investigated magmatic magnetite-hosted car-
bonatites, and experimental data are lacking for the partitioning of
trace elements between magnetite and carbonatite melt. In sulfur-sa-
turated systems, such as Phalaborwa, Cu, V and Co anomalies are not
identified in the multielement diagram (Milani et al., 2017).
The positive correlation between Zn and Mn is distinctive for
magnetite in carbonatite-related environments (Fig. 5c), which is also
recorded in magnetite from the Kerimasi calciocarbonatite from Tan-
zania (Reguir et al., 2008). This coupled partitioning has previously
been reported for magnetite-franklinite aggregates (Sclar and Leonard,
1992). Reguir et al. (2008) suggested that the coupled substitution of
Mn and Zn for iron in the octahedral and tetrahedral sites in the
magnetite structure can minimize the lattice strain induced by the
37
Ore Geology Reviews 107 (2019) 30–40
incorporation of Mn3+ substituting for Fe3+. Although the mechanism
of this covariant partitioning of Zn and Mn is not well understood, it can
be used as a distinctive signature for magnetite formed in carbonatite-
related environments.
5.2. Magnetite chemistry as an ore genetic indicator
A number of researchers have used a variety of diagrams to dis-
tinguish magnetite from different ore genetic environments, in parti-
cular, between magmatic and hydrothermal origins (e.g., Dupuis and
Beaudoin, 2011; Dare et al., 2012; Nadoll et al., 2014; Huang et al.,
2015). The most popularly used discriminant diagrams include: 1) the
Ni + Cr vs. Si + Mg diagram, which separates Ni-Cu-PGE and Cr de-
posits (Dupuis and Beaudoin, 2011); 2) the Ca + Al + Mn vs. Ti + V
and Ni/(Cr + Mn) vs. Ti + V diagrams, which can be used to dis-
criminate between a wide variety of ore deposits, including Iron Oxide
Copper-Gold (IOCG), Kiruna, porphyry Cu, skarn, BIF and Fe-Ti-V de-
posits (Dupuis and Beaudoin, 2011; Nadoll et al., 2014, 2015); 3) the
Al/(Zn + Ca) vs. Cu/(Si + Ca) diagram, which separates VMS deposits
from other deposit types (Dupuis and Beaudoin, 2011); 4) the Ti vs. Ni/
Cr diagram, which distinguishes magnetite forming in hydrothermal
from magmatic settings (Dare et al., 2014); and 5) the Sn/Ga vs. Al/Co
diagram, which discriminates between different magmatic to hydro-
thermal ore deposits (Kamvong et al., 2007).
Our data show that elemental ratios in magnetite are useful for
discriminating magnetite from different ore-forming processes, espe-
cially hydrothermal and magmatic ones, as shown in the following
sections. Elements such as Ti and other HFSEs are generally immobile,
and temperature plays the dominant role in their incorporation in
magnetite (Toplis and Carroll, 1995; Nielsen and Beard, 2000; Toplis
and Corgne, 2002). Thus, magmatic magnetite generally contains
higher Ti concentrations than hydrothermal ones (e.g., Dupuis and
Beaudoin, 2011; Dare et al., 2014; Nadoll et al., 2015; Milani et al.,
2017). Magnetite of magmatic origin in carbonatite-related environ-
ments (e.g., Oka, Mushgai Khudag, Kerimasi, Phalaborwa) is commonly
Ti-rich (> 1 wt%), whereas that with hydrothermal origins (e.g., Hon-
gcheon, Bayan Obo), is generally low in Ti (< 1 wt%; Supplementary
Table 1; Reguir et al., 2008; Milani et al., 2017;). The hydrothermal
magnetites from Hongcheon and Bayan Obo are composed of similar
geochemical features, with typical hydrothermal origins from IOCG,
BIF, skarn and porphyry sources in the multielement diagram (Fig. 8b;
Dare et al., 2014). The hydrothermal magnetite from Hongcheon and
Bayan Obo are more depleted in HFSEs (e.g., Ta, Zr, Hf, V and Ti) re-
lative to the magmatic magnetite from Mushgai Khudag and Oka
(Fig. 8), which is consistent with magnetite in sulfide and silicate sys-
tems (e.g., Dare et al., 2014). Thus, based on the composition of HFSEs,
the discriminant diagrams of Ti vs. Zr + Hf and Ti vs. Nb + Ta have
been proposed to distinguish magmatic from hydrothermal magnetite
within carbonatite complexes (Fig. 9a,b).
W. Chen, et al.
Fig. 9. Plots of Ti vs. Zr + Hf (a), Ti vs. Nb + Ta (b) and Ni/Cr vs. Ti (c) to
distinguish magmatic and hydrothermal magnetite. The boundary (black da-
shed line) refers to Dare et al. (2014).
Dare et al. (2014) found that the behaviors of Ni and Cr are different
in magmatic and hydrothermal magnetite, possibly due to the higher
solubility of Ni in fluids compared to Cr. Hydrothermal magnetite from
carbonatite complexes (Bayan Obo and Hongcheon) generally has Ni/
Cr values larger than 1, whereas magmatic magnetite (Oka and
Mushgai Khudag) shows various Ni/Cr ratios with most values < 1.
Based on these data and their distinct Ti abundances, magmatic and
hydrothermal magnetite from carbonatite complexes can be dis-
tinguished in the plot of Ti vs Ni/Cr (Fig. 9c).
Fig. 10 illustrates that magmatic magnetites from Mushgai Khudag
and Oka both plot within the Fe-Ti-V fields in the discriminant dia-
grams of Al + Mn vs. Ti + V and Ni/(Cr + Mn) vs. Ti + V, as proposed
by Dupuis and Beaudoin (2011). The magmatic magnetite from the
Kerimasi calciocarbonatite also plots in the Fe-Ti-V field in the Al + Mn
vs. Ti + V diagram (Fig. 10a), except for a few magnetite grains, which
are relatively depleted in Ti, that plot close to the boundary of this field.
38
Ore Geology Reviews 107 (2019) 30–40
Fig. 10. Plots of Al + Mn vs. Ti + V (a) and Ni/(Cr + Mn) vs. Ti + V (b) with
reference fields taken from Dupuis and Beaudoin (2011); discrimination dia-
grams of Cu/Mo vs. Zn/Co to distinguish carbonatite-related magnetite from
those within sulfide and silicate systems.
The magnetites from Hongcheon and Bayan Obo plot in the field for
hydrothermal magnetite, from skarn, although a few grains also plot in
the IOCG field (Fig. 10a,b). This result suggests that the discriminant
diagrams proposed by Dupuis and Beaudoin (2011) are effective in the
separation of magmatic and hydrothermal magnetite associated with
carbonatite but not effective in the separation of magnetite formed
within carbonatite environments from other systems. The multielement
diagram (Fig. 8) indicates that elements such as Cu, Zr, Hf, Zn, and Co
can be used to identify carbonatite-related magnetite. As a result, dis-
criminant diagrams using the elemental ratios of Zn/Co and Cu/Mo
shown in Fig. 11a, separates carbonatite-related magnetite from that
from the other environments including Fe-Ti-P deposit, andesite, por-
phyry, IOCG, Fe-skarn and BIF because of its higher Zn/Co and lower
Cu/Mo (Dare et al., 2014; Nadoll et al., 2015). The diagram of Cu vs.
Zr + Hf (Fig. 11b) can also be used to discriminate hydrothermal
magnetite from carbonatite complexes from magnetite within por-
phyry, BIF, Fe-skarn, albitized granitoid and carbonate alteration
complexes (Dare et al., 2014; Wen et al., 2017). Of note, in sulfur sa-
turated carbonatite systems (e.g., Phalaborwa), the discriminant dia-
grams of Zn/Co vs. Cu/Mo and Cu vs. Zr + Hf are note applicable due
to much higher Cu and Co contents in the bulk rock and magnetite
(Milani et al., 2017).
6. Conclusions
In this study, the major, minor and trace element compositions of
W. Chen, et al.
Fig. 11. (a) and Zr + Hf vs. Cu to distinguish hydrothermal magnetite within
carbonatites from other low-temperature magnetite (b). Data for magnetite
from other environments within (a) and (b) are taken from Canil et al. (2016),
Dare et al. (2014) and Wen et al. (2017).
magnetite were obtained using LA-ICP-MS with multiple external re-
ferences without an internal standard. The determined chemical com-
positions of magmatic and hydrothermal magnetite formed from car-
bonatite complexes are quite distinguishable when compared to those
formed in silicate and sulfide melts, as shown in the multielement
variation diagrams. The distinct chemical signatures of carbonatite-re-
lated magnetite include strong positive Mn and Zn anomalies and ne-
gative Cu, Co and Ga anomalies in bulk continental crust-normalized
multielement variation diagrams. This study proposes new discriminant
diagrams of Zn/Co vs. Cu/Mo and Cu vs. Zr + Hf to distinguish car-
bonatite-related magnetite from magnetite formed in other environ-
ments such as Fe-Ti-P deposit, andesite, IOCG, porphyry, BIF, Fe-skarn,
albitized granitoid and carbonate alteration complexes. In addition, the
Ti and HFSE contents in magnetite are useful for distinguishing mag-
matic magnetite from hydrothermal magnetite from carbonatite com-
plexes, as shown in the diagrams of Ti vs. Zr + Hf and Ti vs. Nb + Ta.
Acknowledgements
We appreciate the constructive reviews from three anonymous re-
viewers and the editorial work of editor Franco Pirajno and guest editor
Ignacio Gonzalez-Alvarez. This study is financially supported by the
National Key R&D Program of China (No. 2017YFC0602405), the
National Natural Science Foundation of China (No. 41530211,
41673035), the Fundamental Research Funds for the Central
Universities (CUGCJ1709) and the special fund from the State Key
Laboratory of Geological Processes and Mineral Resources (No.
39
Ore Geology Reviews 107 (2019) 30–40
MSFGPMR03-2).
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.oregeorev.2019.01.029.
References
Baatar, M., Ochir, G., Kynicky, J., Iizumi, S., 2013. Some notes on the Lugiin Gol, Mushgai
Khudag and Bayan Khoshuu alkaline complexes, Southern Mongolia. Int. J. Geosci. 4,
1200–1214.
Bailey, D.K., Kearns, S., 2002. High-Ti magnetite in some fine-grained carbonatites and
the magmatic implications. Mineral. Mag. 66, 379–384.
Barnes, S.J., Maier, W.D., Ashwal, L.D., 2004. Platinum-group element distribution in the
Main Zone and Upper Zone of the Bushveld Complex, South Africa. Chem. Geol. 208,
293–317.
Bell, K., Simonetti, A., 2010. Source of parental melts to carbonatites-critical isotopic
constraints. Mineral. Petrol. 98, 77–89.
Buddington, A.F., Lindsley, D.H., 1964. Iron-Titanium oxide minerals and synthetic
equivalents. J. Petrol. 5, 310–357.
Canil, D., Grondahl, C., Lacourse, T., Pisiak, L.K., 2016. Trace elements in magnetite from
porphyry Cu-Mo-Au deposits in British Columbia, Canada. Ore Geol. Rev. 72,
1116–1128.
Chakhmouradian, A.R., Zaitsev, A.N., 2012. Rare earth mineralization in igneous rocks:
sources and processes. Elements 8, 347–354.
Chen, W., Simonetti, A., 2013. In-situ determination of major and trace elements in
calcite and apatite, and U-Pb ages of apatite from the Oka carbonatite complex: in-
sights into a complex crystallization history. Chem. Geol. 353, 151–172.
Chen, W., Kamenetsky, V.S., Simonetti, A., 2013a. Evidence for the alkaline nature of
parental carbonatite melts at Oka complex in Canada. Nat. Commun. 4, 2687.
Chen, W., Simonetti, A., 2014. Evidence for the multi-stage petrogenetic history of the
Oka carbonatite complex (Quebec, Canada) as recorded by perovskite and apatite.
Minerals 4, 437–476.
Chen, W.T., Zhou, M.F., Zhao, T.P., 2013b. Differentiation of nelsonitic magmas in the
formation of the ∼1.74 Ga Damiao Fe-Ti-P ore deposit, North China. Contrib.
Mineral. Petrol. 165, 1341–1362.
Dare, S.A.S., Barnes, S.J., Beaudoin, G., 2012. Variation in trace element content of
magnetite crystallized from a fractionating sulfide liquid, Sudbury, Canada: im-
plications for provenance discrimination. Geochim. Cosmochim. Acta 88, 27–50.
Dare, S.A.S., Barnes, S.J., Beaudoin, G., Meric, J., Boutroy, E., Potvin-Doucet, C., 2014.
Trace elements in magnetite as petrogenetic indicators. Miner. Deposita 49, 785–796.
Dupuis, C., Beaudoin, G., 2011. Discriminant diagrams for iron oxide trace element fin-
gerprinting of mineral deposit types. Miner. Deposita 46, 319–335.
Enkhbayar, D., Seo, J., Choi, S.G., Lee, Y.J., Batmunkh, E., 2016. Mineral chemistry of
REE-rich apatite and sulfur-rich monazite from the Mushgai Khudag, alkaline vol-
canic-plutonic complex, South Mongolia. Int. J. Geosci. 7, 20–31.
Fan, H.R., Yang, K.F., Hu, F.F., Liu, S., Wang, K.Y., 2016. The giant Bayan Obo REE-Nb-Fe
deposit, China: controversy and ore genesis. Geosci. Front. 7, 335–344.
Gaspar, J.C., Wyllie, P.J., 1983. Magnetite in the carbonatites from the Jacupiranga
Complex, Brazil. Am. Mineral. 68, 195–213.
Gold, D.P., Eby, G.N., Bell, K., Vallée, M., 1986. Carbonatites, diatremes and ultra-alka-
line rocks in the Oka area, Quebec. Geological Association of Canada Guidebook, 21.
Guzmics, T., Mitchell, R.H., Szabo, C., Berkesi, M., Mike, R., Abart, R., 2011. Carbonatite
melt inclusions in coexisting magnetite, apatite and monticellite in Kerimasi calcio-
carbonatite, Tanzania: melt evolution and petrogenesis. Contrib. Miner. Petrol. 161,
177–196.
Hu, Z.C., Gao, S., Liu, Y.S., Hu, S.H., Chen, H.H., Yuan, H.L., 2008. Signal enhancement in
laser ablation ICP-MS by addition of nitrogen in the central channel gas. J. Anal.
Atom. Spectrom. 23, 1093–1101.
Huang, X.W., Zhou, M.F., Qiu, Y.Z., Qi, L., 2015. In-situ LA-ICP-MS trace elemental
analyses of magnetite: the Bayan Obo Fe-REE-Nb deposit, North China. Ore Geol.
Rev. 65, 884–899.
Ivanyuk, G.Y., Kalashnikov, A.O., Pakhomovsky, Y.A., Mikhailova, J.A., Yakovenchuk,
V.N., Konopleva, N.G., Sokharev, V.A., Bazai, A.V., Goryainov, P.M., 2016. Economic
minerals of the Kovdorbaddeleyite-apatite-magnetite deposit, Russia: mineralogy,
spatial distribution and ore processing optimization. Ore Geol. Rev. 77, 279–311.
Kamvong, T., Zaw, K., Siegele, R., 2007. PIXE/PIGE microanalysis of trace elements in
hydrothermal magnetite and exploration significance: a pilot study. 15th Australian
Conference on Nuclear and Complementary Techniques of Analysis and 9th Vacuum
Society of Australia Congress. University of Melbourne, Melbourne, Australia.
Kim, N., Cheong, A.C., Yi, K., Jeong, Y.J., Koh, S.M., 2016. Post-collisional carbonatite-
hosted rare earth element mineralization in the Hongcheon area, central Gyeonggi
massif, Korea: ion microprobe monazite U-Th-Pb geochronology and Nd-Sr isotope
geochemistry. Ore Geol. Rev. 79, 78–87.
Kim, S.J., Lee, H.K., Yin, J.W., Park, J.K., 2005. Chemistry and origin of monazites from
carbonatite dikes in the Hongcheon-Jaeun district, Korea. J. Asian. Earth. Sci. 25,
57–67.
Lattard, D., 1995. Experimental evidence for the exsolution of ilmenite from titaniferous
spinel. Am. Mineral. 80, 968–981.
Lee, M.J., Lee, J.I., Moutte, J., 2005. Compositional variation of Fe-Ti oxides from the
Sokli complex, north-eastern Finland. Geosci. J. 9, 1–13.
Liu, Y.S., Hu, Z.C., Gao, S., Günther, D., Xu, J., Gao, C.G., 2008. In situ analysis of major
W. Chen, et al.
and trace elements of anhydrous minerals by LA-ICP-MS without applying an internal
standard. Chem. Geol. 257, 34–43.
Liu, P.P., Zhou, M.F., Chen, W.T., Gao, J.F., Huang, X.W., 2015. In-situ LA-ICP-MS trace
elemental analyses of magnetite: Fe-Ti-(V) oxide-bearing mafic-ultramafic layered
intrusions of the Emeishan Large Igneous Province, SW China. Ore Geol. Rev. 65,
853–871.
McDonough, W.F., Sun, S.S., 1995. The composition of the Earth. Chem. Geol. 120,
223–253.
Mcmahon, B.M., Haggertry, S.E., 1982. The Oka carbonatite complex: magnetite com-
positions and the related role of Titanium in pyrochlore. In: Meyer, H.O.A., Boyd, F.R.
(Eds.), Kimberlites, diatremes, and diamonds: their geology, petrology and geo-
chemisty 1. American Geophysical Union.
Milani, L., Bolhar, R., Cawthorn, R.G., Frei, D., 2017. In situ LA-ICP-MS and EPMA trace
element characterization of Fe-Ti oxides from the phoscorite-carbonatite association
at Phalaborwa, South Africa. Miner. Deposita 52, 747–768.
Mollo, S., Putirka, K., Iezzi, G., Scarlato, P., 2013. The control of cooling rate on tita-
nomagnetite composition: implications for a geospeedometry model applicable to
alkaline rocks from Mt. Etna volcano. Contrib. Mineral. Petrol. 165, 457–475.
Nadoll, P., Angerer, T., Mauk, J.I., French, D., Walshe, J., 2014. The chemistry of hy-
drothermal magnetite: a review. Ore Geol. Rev. 61, 1–32.
Nadoll, P., Koenig, A.E., 2011. LA-ICP-MS of magnetite: methods and reference materials.
J. Anal. Atom. Spectrom. 26, 1872–1877.
Nadoll, P., Mauk, J.L., Leveille, R.A., Koeing, A.E., 2015. Geochemistry of magnetite from
porphyry Cu and skarn deposits in the southwestern United States. Miner. Deposita
50, 493–515.
Nielsen, R.L., Beard, J.S., 2000. Magnetite-melt HFSE partitioning. Chem. Geol. 164,
21–34.
Prins, P., 1972. Composition of magnetite from carbonatites. Lithos 5, 227–240.
Ramasamy, R., Gwalani, L.G., Subramanian, S.P., 2001. A note on the occurrence and
formation of magnetite in the carbonatites of Sevvattur, North Arcot district, Tamil
Nadu, Southern India. J. Asian Earth Sci. 19, 297–304.
Ray, G.E., Webster, I., 2007. Geology and chemistry of the low Ti magnetite-bearing Heff
Cu-Au skarn and its associated plutonic rocks, Heffley Lake, south-central British
Columbia. Explor. Min. Geol. 16, 159–186.
Reguir, E.P., Chakhmouradian, A.R., Halden, N.M., Yang, P., 2008. Early magmatic and
reaction-induced trends in magnetite from the carbonatites of Kerimasi, Tanzania.
40
Ore Geology Reviews 107 (2019) 30–40
Can. Mineral. 46, 879–900.
Rudnick, R., Gao, S., 2003. Composition of the continental crust. Treatise Geochem. 3,
1–64.
Savard, D., Barnes, S.J., Sunder Raju, P.V., 2010. Accurate LA-ICP-MS calibration for
magnetite analysis using multiple reference materials. Goldschmidt Conference
Abstracts A914.
Sclar, C.B., Leonard, B.F., 1992. Quantitative chemical relationships in a franklinite-
magnetite exsolution intergrowth from Franklin, Sussex County, New Jersey. Econ.
Geol. 87, 1180–1183.
Shin, D., Oh, Y., Lee, M.J., 2013. Mineral chemistry of magnetite from the Hongcheon
carbonatite-phoscorite complex, Korea. J. Miner. Soc. Korea 26, 299–312 (in Korean
with English abstract).
Smith, M.P., Campbell, L.S., Kynicky, J., 2015. A review of the genesis of the world class
Bayan Obo Fe-REE-Nb deposits, Inner Mongolia, China: multistage processes and
outstanding questions. Ore Geol. Rev. 64, 459–476.
Song, W.L., Xu, C., Smith, M.P., Chakmouradian, A.R., Brenna, M., Kynicky, J., Chen, W.,
Yang, Y.H., Deng, M., Tang, H.Y., 2018. Genesis of the world’s largest rare earth
element deposit, Bayan Obo, China: protracted mineralization evolution over ∼1 b.y.
Geology 46, 323–326.
Tan, W., Liu, P., He, H.P., Wang, C.Y., Liang, X.L., 2016. Mineralogy and origin of ex-
solution in Ti-rich magnetite from different magmatic Fe-Ti oxide-bearing intrusions.
Can. Mineral. 54, 539–553.
Toplis, M.J., Carroll, M.R., 1995. An experimental study of the influence of oxygen fu-
gacity on Fe-Ti oxide stability, phase relations, and mineral-melt equilibria in fer-
robasaltic systems. J. Petrol. 36, 1137–1170.
Toplis, M.J., Corgne, A., 2002. An experimental study of element partitioning between
magnetite, clinopyroxene and iron-bearing silicate liquids with particular emphasis
on vanadium. Contrib. Miner. Petrol. 144, 22–37.
Wen, G., Li, J.W., Hofstra, A.H., Koenig, A.E., Lowers, H.A., Adams, D., 2017.
Hydrothermal reequilibration of igneous magnetite in altered granitic plutons and its
implications for magnetite classification schemes: insights from the Handan-Xingtai
iron district, North China Craton. Geochim. Cosmochim. Acta 213, 255–270.
Yang, X., Lai, X., Pirajno, F., Liu, Y., Ling, M., Sun, W., 2017. Genesis of the Bayan Obo Fe-
REE-Nb formation in Inner Mongolia, North China Craton: a perspective review.
Precambr. Res. 288, 39–71.